In one embodiment, edge devices can be configured to be coupled to a multi-stage switch fabric and peripheral processing devices. The edge devices and the multi-stage switch fabric can collectively define a single logical entity. A first edge device from the edge devices can be configured to be coupled to a first peripheral processing device from the peripheral processing devices. The second edge device from the edge devices can be configured to be coupled to a second peripheral processing device from the peripheral processing devices. The first edge device can be configured such that virtual resources including a first virtual resource can be defined at the first peripheral processing device. A network management module coupled to the edge devices and configured to provision the virtual resources such that the first virtual resource can be migrated from the first peripheral processing device to the second peripheral processing device.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. An apparatus, comprising: a multi-stage switch fabric; and a plurality of edge devices having a first plurality of ports configured to be coupled to the multi-stage switch fabric and a second plurality of ports configured to be coupled to a plurality of peripheral processing devices, the plurality of edge devices and the multi-stage switch fabric collectively defining a single logical entity, a first edge device from the plurality of edge devices configured to: classify a data packet, send the data packet through the multi-stage switch fabric based on classification of the data packet, the multi-stage switch fabric having a predictable latency, and implement a congestion resolution scheme configured to isolate the data packet from data packets sent from other edge devices, the multi-stage switch fabric configured not to classify the congestion resolution scheme.
2. The apparatus of claim 1 , wherein the first edge device from the plurality of edge devices is configured to classify the data packet based on at least one of a layer-2 Ethernet address or a layer-4 Ethernet address of the data packet.
This invention relates to network traffic management in edge computing environments, specifically addressing the challenge of efficiently classifying and processing data packets at the network edge. The apparatus includes multiple edge devices that handle data packets by classifying them based on network layer information. The first edge device in the system is configured to classify incoming data packets using either a layer-2 Ethernet address (MAC address) or a layer-4 Ethernet address (transport layer protocol information such as TCP or UDP ports). This classification allows the edge device to determine how to process or route the packet, improving network performance by enabling faster decision-making at the edge. The apparatus may also include additional edge devices that perform further processing or forwarding based on the initial classification. The system optimizes traffic handling by leveraging low-level network identifiers to streamline packet processing, reducing latency and improving efficiency in distributed network architectures. The invention is particularly useful in scenarios requiring real-time data processing, such as IoT networks, content delivery, or cloud-edge computing environments.
3. The apparatus of claim 1 , wherein the first edge device from the plurality of edge devices is configured to: determine a destination of the data packet by classifying the data packet, and send the data packet towards the destination through the multi-stage switch.
This invention relates to a network apparatus for efficiently routing data packets in a multi-stage switch architecture. The problem addressed is the need for intelligent packet classification and routing in edge devices to optimize data flow in large-scale networks, particularly those using multi-stage switches. The apparatus includes a plurality of edge devices connected to a multi-stage switch. Each edge device is configured to classify incoming data packets to determine their destination and then route them through the multi-stage switch. The classification process involves analyzing packet headers or other metadata to identify the appropriate destination, ensuring efficient and accurate routing. The multi-stage switch, which may be a Clos network or similar architecture, provides scalable and high-performance connectivity between edge devices. The edge devices are responsible for both classification and forwarding, reducing the need for centralized control and improving network responsiveness. By distributing the classification task to the edge devices, the system enhances scalability and reduces latency. The multi-stage switch ensures that packets are routed through multiple stages of switching elements, optimizing path selection and load balancing across the network. This approach is particularly useful in data centers, cloud computing environments, and other high-traffic networks where efficient packet routing is critical. The invention improves network performance by minimizing bottlenecks and ensuring that packets are directed to their intended destinations with minimal delay.
4. The apparatus of claim 1 , wherein the first edge device from the plurality of edge device is configured to classify the data packet to determine whether the data packet is an IP packet, a session control protocol packet, a media packet, or a data packet defined at a peripheral processing device.
This invention relates to edge computing systems designed to optimize network traffic processing. The problem addressed is the inefficient handling of diverse data packets in edge networks, where different types of packets (e.g., IP, session control, media, or custom data) require distinct processing paths. The solution involves an apparatus with multiple edge devices, where a first edge device classifies incoming data packets to determine their type. The classification identifies whether the packet is an IP packet, a session control protocol packet, a media packet, or a custom packet defined by a peripheral processing device. This classification enables the system to route packets to appropriate processing modules, improving efficiency and reducing latency. The apparatus may also include additional edge devices that perform further processing based on the classification results. The system ensures that packets are handled according to their specific requirements, enhancing overall network performance. The invention is particularly useful in environments where real-time processing of different packet types is critical, such as in telecommunications, IoT networks, or cloud computing.
5. The apparatus of claim 1 , wherein a number of ports in the second plurality of ports for the plurality of edge devices is at least 1,000.
This invention relates to a high-density port aggregation apparatus designed to efficiently manage and connect a large number of edge devices in a network infrastructure. The apparatus addresses the challenge of scalability in network connectivity, particularly in environments requiring high-density port aggregation to support thousands of edge devices, such as data centers, cloud computing platforms, or large-scale enterprise networks. The apparatus includes a first plurality of ports for connecting to a network backbone and a second plurality of ports for connecting to edge devices. The second plurality of ports is specifically configured to support at least 1,000 edge device connections, enabling high-density aggregation. The apparatus may also include a switching fabric or routing logic to manage data traffic between the backbone and edge devices, ensuring efficient data flow and minimizing latency. Additionally, the apparatus may incorporate power management features, such as power-over-Ethernet (PoE) support, to provide both data and power to connected edge devices. The design may also include redundancy mechanisms, such as failover ports or load-balancing logic, to enhance reliability and uptime. The apparatus is optimized for high-density environments, reducing physical footprint while maximizing port density and performance.
6. The apparatus of claim 1 , wherein a number of ports in the second plurality of ports for the plurality of edge devices is at least 100,000.
This invention relates to high-density networking apparatus designed to support a large number of edge devices, addressing the challenge of efficiently managing and connecting numerous endpoints in data centers or telecommunications networks. The apparatus includes a first plurality of ports for connecting to a network backbone and a second plurality of ports for connecting to edge devices, such as user terminals or IoT devices. The second plurality of ports is configured to support at least 100,000 connections, enabling massive scalability for high-density environments. The apparatus may also include switching logic to dynamically route traffic between the backbone and edge devices, ensuring low-latency and high-throughput performance. Additionally, the system may incorporate power-over-data capabilities, allowing simultaneous data transmission and power delivery to connected devices. The design optimizes space and energy efficiency while maintaining reliability, making it suitable for large-scale deployments in cloud computing, smart cities, or industrial automation. The apparatus may further include redundancy features, such as backup power supplies or failover mechanisms, to enhance operational resilience. By supporting a high port density, this invention simplifies network infrastructure by reducing the need for multiple intermediate switches, thereby lowering costs and complexity.
7. The apparatus of claim 1 , wherein the plurality of peripheral processing devices include at least one of (1) a plurality of compute nodes, (2) a plurality of service nodes, (3) a plurality of routers, or (4) a plurality of storage nodes.
This invention relates to a distributed computing system designed to enhance scalability and resource management. The system includes a central controller and multiple peripheral processing devices, which may consist of compute nodes, service nodes, routers, or storage nodes. These peripheral devices are interconnected and managed by the central controller to optimize task distribution, data routing, and resource allocation. The central controller dynamically assigns tasks to the peripheral devices based on their capabilities and current workload, ensuring efficient utilization of system resources. The system is particularly useful in high-performance computing environments where tasks must be processed in parallel across multiple nodes while maintaining low latency and high throughput. The inclusion of routers facilitates seamless communication between nodes, while storage nodes provide scalable data storage. Service nodes handle specialized functions such as authentication, monitoring, or load balancing. The overall architecture allows for flexible expansion, enabling the addition of more nodes as demand grows. This design addresses challenges in distributed computing, such as load balancing, fault tolerance, and efficient resource allocation, by providing a modular and adaptable framework.
8. The apparatus of claim 1 , wherein the multi-stage switch fabric is configured to redirect the data packet based on a switch table.
A multi-stage switch fabric apparatus is designed to efficiently route data packets within a network. The apparatus addresses the challenge of optimizing data flow in high-performance networking environments, where traditional switching methods may introduce latency or bottlenecks. The switch fabric includes multiple stages of switching elements that collaboratively process and forward data packets. Each stage operates in sequence, applying routing decisions to ensure packets reach their intended destinations with minimal delay. The apparatus further includes a switch table that stores routing information, enabling the multi-stage switch fabric to dynamically redirect data packets based on predefined rules or real-time conditions. The switch table may contain entries that map packet attributes, such as source and destination addresses, to specific output ports or paths within the fabric. By consulting this table, the switch fabric can make intelligent routing decisions, improving network efficiency and reducing congestion. The multi-stage architecture allows for scalable and flexible packet routing, accommodating varying network loads and topologies. The switch fabric may also support parallel processing of packets across multiple stages, enhancing throughput and reducing latency. Additionally, the apparatus may include mechanisms for error detection and correction, ensuring reliable data transmission. The overall design aims to provide a high-performance, adaptable switching solution for modern networking applications.
9. The apparatus of claim 1 , wherein the first edge device from the plurality of edge device is further configured to: parse the data packet into a plurality of cells, and concatenate switching information to each cell in the plurality of cells, the switching information including at least one of header information, destination information, or source information associated with each cell in the plurality of cells.
This invention relates to edge computing systems, specifically improving data packet processing in distributed networks. The problem addressed is inefficient data handling in edge networks, where traditional packet-based routing can introduce latency and complexity. The solution involves an apparatus with multiple edge devices that enhance data transmission by parsing packets into smaller cells and attaching switching information to each cell. This switching information includes header data, destination details, and source identifiers, enabling faster and more flexible routing decisions. The apparatus ensures that each cell carries sufficient metadata for independent processing, allowing distributed edge nodes to handle traffic more efficiently. By breaking down packets into cells and enriching them with context-specific switching data, the system reduces overhead and improves scalability in edge networks. The approach is particularly useful in scenarios requiring low-latency communication, such as IoT deployments or real-time applications, where traditional packet switching may not suffice. The invention optimizes data flow by leveraging cell-based segmentation and metadata augmentation, streamlining operations across edge devices.
10. The apparatus of claim 1 , wherein the first edge device from the plurality of edge devices is further configured to: perform a scheduling of transmission of the data packet before sending the data packet, the multi-stage switch fabric is configured not to perform the scheduling of transmission.
Edge computing systems often face challenges in efficiently routing data packets between edge devices and central processing units (CPUs) through multi-stage switch fabrics. Traditional approaches may rely on the switch fabric to handle scheduling, which can introduce latency and reduce overall system performance. This invention addresses these issues by implementing scheduling at the edge device level rather than within the switch fabric. The apparatus includes a plurality of edge devices interconnected via a multi-stage switch fabric, where the switch fabric is designed without scheduling capabilities. Instead, each edge device is configured to perform scheduling of data packet transmission before sending the packet. This pre-scheduling ensures that the switch fabric operates more efficiently by receiving already scheduled packets, reducing latency and improving throughput. The edge devices may also include additional features such as packet processing, routing, and prioritization to optimize data flow. The multi-stage switch fabric facilitates high-speed, low-latency communication between edge devices and CPUs, leveraging the pre-scheduled packets to minimize delays. This approach enhances system performance by offloading scheduling tasks from the switch fabric, allowing it to focus on data forwarding. The invention is particularly useful in high-performance computing environments where low latency and high throughput are critical.
11. A method, comprising: receiving a data packet at a first edge device from a second edge device from a plurality of edge devices, the plurality of edge devices having a first plurality of ports configured to be coupled to a multi-stage switch fabric and a second plurality of ports configured to be coupled to a plurality of peripheral processing devices, the plurality of edge devices and the multi-stage switch fabric collectively defining a single logical entity; classifying, at the first edge device, the data packet; and sending, from the first edge device, the data packet through the multi-stage switch fabric based on classification of the data packet; implementing a congestion resolution scheme to isolate the data packet from data packets sent from edge devices in the plurality of edge devices other than the first edge device and the second edge device.
This invention relates to data packet routing in a network architecture involving edge devices and a multi-stage switch fabric. The system addresses the challenge of efficiently managing data traffic while preventing congestion and ensuring proper packet classification and routing. The architecture includes multiple edge devices, each with two sets of ports: one connected to a multi-stage switch fabric and another connected to peripheral processing devices. These components collectively form a single logical entity, enabling seamless data flow. When a data packet is received at a first edge device from a second edge device, the packet is classified based on predefined criteria. The classified packet is then routed through the multi-stage switch fabric. To mitigate congestion, the system implements a congestion resolution scheme that isolates the packet from others originating from different edge devices, ensuring that only packets from the first and second edge devices are considered in the routing decision. This approach optimizes network performance by preventing bottlenecks and maintaining efficient data transmission.
12. The method of claim 11 , wherein classifying the data packet includes classifying the data packet based on at least one of a layer-2 Ethernet address or a layer-4 Ethernet address of the data packet.
A method for classifying data packets in a network involves analyzing packet characteristics to determine their type or origin. The method includes classifying data packets based on at least one of a layer-2 Ethernet address or a layer-4 Ethernet address within the packet. Layer-2 classification involves examining the MAC (Media Access Control) address, which identifies devices on a local network segment. Layer-4 classification involves analyzing transport layer information, such as TCP or UDP port numbers, which help identify the application or service associated with the packet. By using these addresses, the method can categorize packets for routing, security filtering, or traffic management. The classification process may also involve comparing the extracted addresses against predefined rules or databases to determine the packet's origin, destination, or intended use. This approach enhances network efficiency by enabling precise packet handling based on their characteristics. The method is particularly useful in environments where accurate classification is critical, such as in cybersecurity, network monitoring, or quality of service (QoS) management.
13. The method of claim 11 , wherein classifying the data packet includes determining a destination of the data packet and sending the data packet through the multi-stage switch fabric includes sending the data packet towards the destination through the multi-stage switch.
A method for efficiently routing data packets in a multi-stage switch fabric network addresses the challenge of optimizing packet forwarding in high-performance computing and data center environments. The method involves classifying data packets based on their destination and dynamically routing them through a multi-stage switch fabric to minimize latency and maximize throughput. The classification process determines the intended destination of each packet, which may include a specific server, storage device, or network endpoint. Once classified, the packet is forwarded through the multi-stage switch fabric, which consists of multiple interconnected switching stages designed to handle high-speed data traffic. The routing process ensures that packets are directed toward their destination with minimal hops, reducing congestion and improving overall network efficiency. This approach is particularly useful in large-scale networks where traditional switching architectures may struggle to maintain low latency and high throughput. The method leverages the multi-stage switch fabric's ability to distribute traffic across multiple paths, enhancing reliability and scalability. By dynamically adjusting routing decisions based on real-time network conditions, the system adapts to varying traffic patterns, ensuring consistent performance. This technique is applicable in high-performance computing clusters, cloud data centers, and other environments requiring efficient data packet routing.
14. The method of claim 11 , wherein classifying the data packet includes determining whether the data packet is an IP packet, a session control protocol packet, a media packet, or a data packet defined at a peripheral processing device.
A method for classifying data packets in a network processing system addresses the challenge of efficiently identifying and categorizing different types of network traffic to optimize routing, security, and processing. The method involves analyzing incoming data packets to determine their type, which may include IP packets, session control protocol packets, media packets, or data packets defined by a peripheral processing device. IP packets are standard internet protocol packets used for general data transmission. Session control protocol packets are used for establishing, managing, and terminating communication sessions, such as those in VoIP or video conferencing. Media packets carry real-time audio, video, or multimedia content. Peripheral processing devices may define custom data packets specific to their functions, such as industrial control systems or specialized sensors. By accurately classifying these packets, the system can apply appropriate handling rules, such as prioritizing media packets for low-latency processing or routing session control packets to security inspection modules. This classification improves network efficiency, reduces latency, and enhances security by ensuring packets are processed according to their specific requirements. The method supports dynamic adaptation to different network environments and protocols, making it suitable for diverse applications in telecommunications, IoT, and enterprise networks.
15. The method of claim 11 , wherein the multi-stage switch fabric is configured to redirect the data packet based on a switch table.
A multi-stage switch fabric system is designed to efficiently route data packets within a network infrastructure. The system addresses the challenge of optimizing packet forwarding in high-performance networks by dynamically adjusting routing paths to improve throughput and reduce latency. The switch fabric includes multiple stages of switching elements that collaboratively process and forward data packets. Each stage applies specific routing logic to determine the optimal path for packet transmission, ensuring efficient data flow across the network. The system incorporates a switch table that stores routing information, allowing the switch fabric to make intelligent forwarding decisions. The switch table contains entries that map destination addresses to corresponding output ports, enabling the switch fabric to redirect data packets based on predefined or dynamically updated routing rules. This mechanism enhances network adaptability and performance by minimizing congestion and optimizing resource utilization. The multi-stage architecture of the switch fabric allows for scalable and flexible packet routing. Each stage of the switch fabric can independently process packets, applying different routing criteria as needed. This modular design supports high-speed data transmission and ensures reliable packet delivery even under varying network conditions. The system is particularly useful in large-scale networks where efficient packet forwarding is critical for maintaining performance and reliability.
16. The method of claim 11 , further comprising: parsing the data packet into a plurality of cells at the first edge device; and concatenating switching information to each cell in the plurality of cells, the switching information including at least one of header information, destination information, or source information associated with each cell in the plurality of cells.
This invention relates to data packet processing in network communication systems, specifically addressing the challenge of efficiently routing data packets through edge devices in a network. The method involves parsing a data packet into multiple smaller cells at a first edge device, which facilitates more granular handling of the data. Each cell is then augmented with switching information, including header information, destination information, or source information, to enable proper routing and processing within the network. This approach improves data transmission efficiency by allowing individual cells to be independently routed and reassembled at their destination, reducing latency and enhancing reliability in high-traffic or complex network environments. The switching information ensures that each cell retains context about its origin and intended path, supporting seamless integration with existing network protocols and infrastructure. The method is particularly useful in scenarios requiring high-speed, low-latency data transfer, such as cloud computing, telecommunication networks, or distributed computing systems. By breaking down packets into cells and attaching detailed routing metadata, the invention optimizes network performance while maintaining data integrity and reducing the risk of packet loss or misrouting.
17. The method of claim 11 , wherein the multi-stage switch fabric is configured not to classify the data packet and not to classify the congestion resolution scheme.
A multi-stage switch fabric is used in high-speed networking to efficiently route data packets between multiple input and output ports. A key challenge in such systems is managing packet classification and congestion resolution, which can introduce latency and complexity. This invention addresses this by simplifying the switch fabric's operation. The switch fabric is designed to avoid classifying data packets, meaning it does not analyze or categorize packets based on their content, headers, or other attributes. Additionally, it does not classify congestion resolution schemes, meaning it does not select or apply different congestion control methods based on packet characteristics or network conditions. Instead, the switch fabric relies on predefined, uniform handling of all packets and congestion scenarios, reducing processing overhead and improving throughput. This approach is particularly useful in high-performance networking environments where low latency and high efficiency are critical. The switch fabric may include multiple stages, such as ingress, fabric, and egress stages, each contributing to the overall routing process without performing classification tasks. By eliminating classification steps, the system achieves faster packet forwarding and simpler hardware design.
18. The method of claim 11 , further comprising: performing, at the first edge device, a scheduling of transmission of the data packet before sending the data packet, the multi-stage switch fabric configured not to perform the scheduling of transmission.
This invention relates to data transmission in a networked system using a multi-stage switch fabric. The problem addressed is the inefficiency in data packet transmission when scheduling is not performed at the edge devices, leading to potential congestion or delays in the network. The solution involves performing transmission scheduling at the first edge device before sending the data packet, while the multi-stage switch fabric itself does not handle this scheduling. This ensures that data packets are transmitted in an optimized manner, reducing congestion and improving overall network performance. The edge device, which is part of the network and responsible for initial data handling, schedules the transmission of the data packet based on network conditions, priority, or other factors. The multi-stage switch fabric, which is a network infrastructure component that routes data packets through multiple stages of switches, relies on the pre-scheduled data packets from the edge device to maintain efficient data flow. This approach decentralizes scheduling, allowing the edge devices to manage transmission timing while the switch fabric focuses on routing. The result is a more balanced and efficient data transmission system.
19. An apparatus, comprising: a multi-stage switch fabric; and a plurality of edge devices having a first plurality of ports configured to be coupled to the multi-stage switch fabric and a second plurality of ports configured to be coupled to a plurality of peripheral processing devices, a number of ports in the second plurality of ports for the plurality of edge devices being at least 10,000, the plurality of edge devices and the multi-stage switch fabric collectively defining a single logical entity, a first edge device from the plurality of edge devices configured to: classify a data packet received from a second edge device in the plurality of edge devices, parse the data packet into a plurality of cells, concatenate switching information to each cell in the plurality of cells, the switching information includes at least one of header information, destination information, or source information associated with each cell in the plurality of cells, send the plurality of cells through the multi-stage switch fabric based on classification of the data packet, the multi-stage switch fabric is configured to redirect the plurality of cells based on a switch table, and implement a congestion resolution scheme to isolate the data packet from data packets sent from edge devices in the plurality of edge devices other than the first edge device and the second edge device.
This invention relates to a high-performance networking apparatus designed to handle large-scale data traffic efficiently. The apparatus includes a multi-stage switch fabric and multiple edge devices, each with at least 10,000 ports for connecting to peripheral processing devices. The edge devices and switch fabric operate as a unified logical entity, enabling seamless data routing. When an edge device receives a data packet from another edge device, it classifies the packet, parses it into smaller cells, and appends switching information such as headers, destination, and source details to each cell. These cells are then transmitted through the multi-stage switch fabric, which uses a switch table to redirect them while implementing congestion control to isolate the packet from other traffic. This isolation prevents interference between different data flows, ensuring reliable and efficient data transmission in high-density networking environments. The system is particularly suited for large-scale data centers or distributed computing systems requiring high throughput and low latency.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
September 18, 2019
March 8, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.